From OpenSCADAWiki
- * — includes of double call the gettimeofday() function.
- ** — pow(Pi,2) measured in an environment with GLibC > 2.31 (like to Debian 11) where performance of the function calculation was degraded on [46...60]%.
- "Entry into JavaLikeCalc" — entry to a procedure on the language JavaLikeCalc means also entry to the critical section and requesting for reading of the RW lock, then the time mostly show performance of the locking operation. The time was excluded from related values in columns with JavaLikeCalc.
- "Operation sin(Pi) and pow(Pi,2)" — after year 7528/2020 and version 0.9.3 LTS the measured value can be risen from 2% to 5% (on slow memory) through a reason of adding control of execution the functions;
- "DDR3-1600-{N}GB/s" — where N is gotten by the program sysbench (not memtest or mbw which are measuring in different methods), by the typical call command sysbench --test=memory --memory-total-size=3G --memory-block-size=1M run for systems with memory size [4...15] GB.
-
The memory test here up to version 1.0 uses other method, which show speed in three times lesser, especially on AMD!
- "precision_test" — a simple program of the OpenSCADA project to test lag of the threads invoking in the RT policy Round-robin and priority 80. 100% loading is performed by the program highload and the VFS activity is performed by the command dd if=/dev/zero of=/var/tmp/test.zero bs=1048576 count=1000; dd if=/var/tmp/test.zero of=/dev/null bs=1048576;.